We have had at least 7 nodes connected to Juniper EX4500 switches (A & B side) for over a year. Recently upgraded disk firmware and upon reboot one of the X200 nodes could not establish client connections for a specific subnet/vlan. The switch drops the bridge entry as fast as it is added. We had to remove the 10G interface from the pool and instead use the 1G interface. We have an open case with Juniper, awaiting some feedback or direction.
Gents, Ask your Cisco Switch admins, to look at spanning tree (assuming that you have 2x10Gbe per node connected). I've seen something like this on Nexus switches before when the admin had changed away from Cisco's default metrics for STP. Effectively one or both of the ports get put into a blocked state, because the switches think that there is another switch attached and start sending over BPDUs to confirm that.
AndrewChung
132 Posts
0
July 19th, 2013 09:00
What type of connection/cabling are you using? Optical or TwinAx? What specific SFPs are you using on each side and the revision?
ShaneCrist
4 Posts
0
March 14th, 2014 07:00
Hi,
OneFS 7.0.2.4
NL & X200 nodes
We have had at least 7 nodes connected to Juniper EX4500 switches (A & B side) for over a year. Recently upgraded disk firmware and upon reboot one of the X200 nodes could not establish client connections for a specific subnet/vlan. The switch drops the bridge entry as fast as it is added. We had to remove the 10G interface from the pool and instead use the 1G interface. We have an open case with Juniper, awaiting some feedback or direction.
crklosterman
450 Posts
0
March 20th, 2014 17:00
Gents, Ask your Cisco Switch admins, to look at spanning tree (assuming that you have 2x10Gbe per node connected). I've seen something like this on Nexus switches before when the admin had changed away from Cisco's default metrics for STP. Effectively one or both of the ports get put into a blocked state, because the switches think that there is another switch attached and start sending over BPDUs to confirm that.
~Chris